Global Change Biology (2012) 18, 902–914, doi: 10.1111/j.1365-2486.2011.02580.x
The effects of land use and climate change on the carbon
cycle of Europe over the past 500 years
J E D O . K A P L A N * , K R I S T E N M . K R U M H A R D T * and N I K L A U S E . Z I M M E R M A N N †
*Soil-Vegetation-Atmosphere Research Group, Institute of Environmental Engineering, Ecole Polytechnique Fédérale de Lausanne,
Station 2, 1015 Lausanne, Switzerland, †Landscape Dynamics Unit, Swiss Federal Research Institute WSL, Zuercherstrasse 111,
8903 Birmensdorf, Switzerland
Abstract
The long residence time of carbon in forests and soils means that both the current state and future behavior of the
terrestrial biosphere are influenced by past variability in climate and anthropogenic land use. Over the last half-millennium, European terrestrial ecosystems were affected by the cool temperatures of the Little Ice Age, rising CO2
concentrations, and human induced deforestation and land abandonment. To quantify the importance of these processes, we performed a series of simulations with the LPJ dynamic vegetation model driven by reconstructed
climate, land use, and CO2 concentrations. Although land use change was the major control on the carbon inventory
of Europe over the last 500 years, the current state of the terrestrial biosphere is largely controlled by land use
change during the past century. Between 1500 and 2000, climate variability led to temporary sequestration events of
up to 3 Pg, whereas increasing atmospheric CO2 concentrations during the 20th century led to an increase in carbon
storage of up to 15 Pg. Anthropogenic land use caused between 25 Pg of carbon emissions and 5 Pg of uptake over
the same time period, depending on the historical and spatial pattern of past land use and the timing of the reversal
from deforestation to afforestation during the last two centuries. None of the currently existing anthropogenic land
use change datasets adequately capture the timing of the forest transition in most European countries as recorded
in historical observations. Despite considerable uncertainty, our scenarios indicate that with limited management,
extant European forests have the potential to absorb between 5 and 12 Pg of carbon at the present day.
Keywords: carbon cycle, carbon sequestration, climate change, dynamic global vegetation model, European history, forest transition, land use
Received 13 July 2011 and accepted 26 September 2011
Introduction
The state of terrestrial ecosystems at the present day is
a product of past climate and anthropogenic land cover
change (ALCC; Pongratz et al., 2009; Strassmann et al.,
2008). Many processes in the terrestrial biosphere operate on centennial to millennial timescales including the
growth and development of forests and the dynamics
of soil organic matter decomposition (Trumbore, 2000;
Canadell et al., 2007). Although there is evidence that
even old trees continue to accumulate carbon at a
steady rate (Luyssaert et al., 2008), forests tend to be
most productive during their early stages of growth; as
a forest ecosystem matures, the rate at which it sequesters carbon tends to decrease (Albani et al., 2006) and
net ecosystem production approaches zero (Canadell
et al., 2007). Thus, the potential for carbon to be stored
in terrestrial ecosystems in the future depends strongly
on past ecosystem history and the trajectory of ecosystems at present (Magnani et al., 2007).
Correspondence: Jed O. Kaplan, tel. + 41 21 693 8058,
fax + 41 21 693 3913, e-mail: [email protected]
902
In Europe, the five centuries following 1500 CE are
distinguished by climate variability, including the Little Ice Age (LIA), and intense human impact on terrestrial ecosystems, mainly through the conversion of
forests to cropland and pasture, and the exploitation
of forests for fuel and construction materials. Following the Industrial Revolution, most European countries experienced a forest transition (Mather et al.,
1998b; Mather, 1999), the afforestation that typically
accompanies societies reaching a certain level of economic development, urbanization and technological
advancements in agriculture. Forest transitions commonly occurred in central, west, and Nordic European
countries from the beginning of the 19th century
(Mather, 1999; Rudel et al., 2005; Bradshaw, 2008) and
later during the 20th century in Eastern Europe
(Mather, 1992; Kauppi et al., 2006; Kozak et al., 2007).
The shift of European ecosystems from source to sink
and subsequent rates of carbon sequestration with
forest regrowth depend on the timing of the initial
forest transition and external environmental factors
including changes in climate and atmospheric CO2
concentrations.
© 2011 Blackwell Publishing Ltd
T H E E F F E C T S O F L A N D U S E A N D C L I M A T E C H A N G E 903
It is generally thought that ecosystems in the temperate latitudes of the northern hemisphere represent
the main natural terrestrial sink for CO2 over recent
decades as a result of land use change, and CO2 and
N fertilization (McGuire et al., 2001; Gurney et al.,
2004; Friedlingstein et al., 2006; Denman et al., 2007).
In Europe, evidence from both forest inventory data
and modeling studies suggests that forests represented
a substantial carbon sink in the late 20th century. In a
synthesis of forest inventory data, Ciais et al. (2008)
observed large increases in forest biomass and net primary production (NPP) during the period from 1950
to 2000. These authors concluded that environmental
factors, such as increasing atmospheric CO2, were the
main reason for the increase in NPP, whereas
increases in biomass were attributed to changes in forest management. In contrast, the model intercomparison study of Churkina et al. (2010) attributed a large
portion of the net carbon uptake in European ecosystems to CO2 fertilization, and suggested that climate
and land cover changes over the second half of the
20th century were either neutral or sources of atmospheric CO2.
Despite the fact that carbon dynamics in terrestrial
ecosystems evolve over much longer timescales than
just a few decades, few studies have tried to take the
multicentennial to millennial view of changes in terrestrial carbon stocks and vegetation productivity on continental scales. Reliable land use and forest inventory
data exists only from the mid-20th century (Ramankutty & Foley, 1999; Klein Goldewijk, 2001; Klein
Goldewijk et al., 2011), so most attempts to quantify
changes in the terrestrial biosphere and the carbon balance of ecosystems as a result of ALCC over longer
time periods have been limited to bookkeeping methods (Houghton, 1999) and/or simple extrapolation of
present day land use patterns to the distant past (Jain &
Yang, 2005; Olofsson & Hickler, 2008).
Recently, a study of ALCC in Europe offered an alternative view of land cover change for this highly
impacted area of the world, depicting much higher levels of preindustrial anthropogenic land use than previous extrapolation-based methods (Fig. 1; Kaplan et al.,
2009). Expanding their method of reconstructing ALCC
to global scale and employing a dynamic global vegetation model, Kaplan et al. (2010) suggested that previous
estimates of Holocene carbon emissions as a result of
ALCC were greatly underestimated. This study did not
account for climate variability however, which could
have had confounding or synergistic impacts on the terrestrial carbon cycle with respect to ALCC (Pongratz
et al., 2009; Jungclaus et al., 2010).
As a result of the very good spatial coverage of
paleoclimate proxy records and historical observa© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
(a)
(c)
(b)
(d)
Fig. 1 Patterns and trends of anthropogenic land use: (a) KK10
scenarios at year 1500, (b) HYDE at year 1500, (c) HYDE at year
2000, and d, time series of land cover change from 1500 to 2000
for the area shown on the maps: KK10-merged in blue, KK10interpolated in orange, and HYDE in red.
tions, the past climate of Europe can be reconstructed
with reasonably high spatial and temporal resolution.
Over the past two millennia, the climate of Europe is
observed to have been generally cooler and more
variable than the 20th century average (e.g., Büntgen
et al., 2011), and also showed important spatial variations (Luterbacher et al., 2004; Pauling et al., 2006).
The LIA, and the following transition to the warmer
20th century concurrent with industrialization and
increases in atmospheric CO2 concentrations could
have had significant effects on European terrestrial
ecosystems and their carbon stocks at the present
day and their potential to sequester additional carbon
in the future.
To quantify the effects of ALCC, climate change, and
increasing CO2 on the carbon cycle of Europe over the
past 500 years we assembled datasets of past climate
and ALCC and performed a number of experiments
and sensitivity tests using the Lund-Potsdam-Jena
dynamic global vegetation model (Sitch et al., 2003). We
place our results into the context of other modeling and
observation-based estimates of the current state of terrestrial ecosystems in Europe, and we discuss the
potential for future carbon sequestration based on the
trajectory of ecosystems at the present and differences
with our control simulations.
904 K A P L A N e t a l .
With regards to the treatment of human impact in
this study, we only consider the affect of conversion
of land from unmanaged landscapes to agricultural
land and pasture, and the abandonment of land used
for these purposes. While we believe that land conversion and abandonment are the most important
anthropogenic processes influencing total carbon in
terrestrial ecosystems, other factors not considered in
this study, were widespread in Europe over the past
five centuries and also influenced carbon stocks and
cycling. These processes include forest harvest and
management (coppicing, pollarding), litter collection
and transfer, soil erosion and degradation, draining
of wet soils and peatlands, air pollution and atmospheric deposition of sulfate, nitrate and metals (e.g.,
Ebermayer, 1876; Ellenberg, 1988; Glatzel, 1991).
Nearly all of the European land area was subject to
these processes in addition to basic conversion and
abandonment, and the subject should be the focus of
future research.
Materials and methods
Model setup
LPJ is driven by gridded soils data and time series of temperature, precipitation, cloud cover, and atmospheric CO2
concentrations. We developed a paleoclimate scenario for
LPJ based on the 500-year high-resolution gridded reconstructions of seasonal temperature and precipitation from
historical observations and biological proxies (Luterbacher
et al., 2004; Pauling et al., 2006). Cloud cover was reconstructed with a bivariate linear model based on temperature
and precipitation at each gridcell. For model runs with fixed
climate, we detrended the CRU TS2.1 observation-based
dataset and added this to the 1961–1990 mean monthly climatology (Mitchell & Jones, 2005). For model runs with
transient atmospheric CO2 concentrations, we used an annually resolved time series of CO2 made by a weighted spline
fit through Antarctic ice-core, firn and observational records
(Krumhardt & Kaplan, 2010). For runs with fixed CO2, the
model was run with 280.6 ppm, the approximate mean global concentration of atmospheric CO2 at year 1500. Soil texture, water holding capacity, and saturated hydraulic
conductivity are from the Harmonized World Soil Database
(FAO/IIASA/ISRIC/ISSCAS/JRC, 2009) combined with the
ISRIC-WISE taxo-transfer database (Batjes, 2006). Our version of LPJ contains an improved observation-based geographic distribution of soil organic matter, where the mean
turnover time of soil organic matter is a function of total soil
clay content, following Kaplan et al. (2010). This modification
generally has the effect of reducing the mean turnover time
of soil organic matter in cool-temperate ecosystems relative
to the original LPJ, particularly in regions with sandy or
young soils, e.g., on areas that were ice-covered during the
last glaciation. LPJ was spun up for 1000 years to reach
equilibrium conditions at 1500 for each model run.
Land use scenarios
The principal ALCC dataset used in this study is based on a
recent study (Kaplan et al., 2009), in which the authors created
a new scenario of prehistoric and preindustrial ALCC in Europe between 1000 BC and 1850. In this study, ALCC was simulated using historical population estimates and a nonlinear
human population-forest area relationship based on historical
land cover records from several European countries. The
resulting dataset depicts substantially higher levels of preindustrial ALCC than works that rely on population-scaled
extrapolation of 20th century trends in ALCC (Klein Goldewijk et al., 2011). This dataset, referred to hereafter as the
KK10 scenario (Kaplan et al., 2010), was modified for this
study in two ways: (1) We improved the distribution of usable
land and thus the spatial distribution of some of the ALCC,
and (2) we extended the datasets to present day using 20th
century data from the HYDE 3.1 historical land use database
(HYDE; Klein Goldewijk et al., 2011).
Land use in the original KK10 dataset was distributed
based on a datasets of land suitability for agriculture and
usable land, i.e., land with a climate suitable enough for any
use by humans. This results in a fairly reasonable distribution
of ALCC, except in northern areas of Europe (e.g., Scotland
and Scandinavia) where land is too cold, and in desert areas
where land was assigned as too dry for land use, although
irrigation is possible (e.g., along the Nile River). To help make
the distribution of usable land more realistic, in addition to the
original dataset of land suitability for agriculture, we used a
new dataset of usable land: a gridded dataset of the maximum
fractional land use ever assigned in HYDE (Klein Goldewijk
et al., 2011). This results in maps of land cover that are similar
to the original Standard scenario from Kaplan et al. (2009), but
with improved ALCC distribution in cold or dry areas. Furthermore, by redistributing ALCC in this manner, we accommodate a smoother merge of the KK10 data, which ends at
1850, to the modern ALCC estimates from the HYDE dataset.
To bring the KK10 land use scenario up to present day, we
developed a strategy to splice the final years of KK10 into the
HYDE data over the period 1850–2000. A description of our
splicing methodology and a conceptual figure (Fig. S1) are
detailed in the supplementary materials. In contrast to this
smooth merging technique, we also included a simple linear
interpolation of the KK10 data at 1850 to HYDE at 2000. To
further quantify the importance of the ALCC dataset used, we
included the standard HYDE dataset (Klein Goldewijk et al.,
2011) as an alternative to the two versions of the KK10
scenario. Therefore, the three ALCC scenarios used in this
study are:
1 KK10 ALCC scenario smoothly merged to modern-day
HYDE data (KK10-merged).
2 KK10 ALCC scenario linearly interpolated from 1850 to
HYDE at year 2000 (KK10-interpolated).
3 HYDE ALCC scenario.
© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
T H E E F F E C T S O F L A N D U S E A N D C L I M A T E C H A N G E 905
The major differences between the HYDE and KK10 scenarios are the amount of land use at 1500 and the subsequent
trends in ALCC (Fig. 1). Their principal similarities, however,
are that they all terminate with the same ALCC fractions of
HYDE at year 2000 and they all have similar maximum levels
of land use in Europe (Fig. 1c and d).
Model runs
To quantify the importance of climate, CO2, and ALCC on the
carbon cycle of Europe, we ran a large series of LPJ model
runs (Table 1). We ran the model with reconstructed paleoclimate for the past 500 years and no ALCC, i.e., natural vegetation change only, fixed climate of the 20th century, fixed CO2,
each of the ALCC scenarios described above, and combinations of the above drivers.
Anthropogenic land cover change was input on an annual
time step and handled with respect to change in carbon using
established methodology (McGuire et al., 2001; Strassmann
et al., 2008; Kaplan et al., 2010). This includes simulating up
to three tiles per gridcell: (1) land that has never been under
land use, (2) land currently used by people, and (3) abandoned land that is recovering from past use. Clearance of natural vegetation results in the transfer of all leaf and
belowground biomass to the litter pool and immediate oxidation of 25% of the above ground woody biomass. The rest of
the woody biomass is transferred to 2- and 20-year exponential decay pools. Once agricultural land, harvest is simulated
by annually oxidizing 100% of the aboveground biomass on
the agricultural tile; upon harvest the belowground living
root biomass is transferred to the belowground litter pool.
Natural fires are the only form of disturbance simulated in
this version of LPJ.
Results
Our model results demonstrate the various impacts of
millennial-scale climate and ALCC, both separately and
combined, on the state of the terrestrial biosphere of
Europe in the year 2000 and the potential for future carbon sequestration. While climate variability resulted in
shorter-term increases in carbon storage, CO2 fertilization substantially increased carbon stocks during the
20th century. Conversely, ALCC in Europe has resulted
in major losses of terrestrial carbon during or prior to
the last 500 years. In the following sections, all changes
in carbon storage are described as cumulative sinks or
sources relative to the amounts simulated at the beginning of the model run in 1500 CE.
Climate and CO2 effects on the terrestrial biosphere
Climate change alone over the past 500 years led to
modest fluctuations in carbon storage, which, prior to
1950, generally resulted in a carbon sink. By comparing
model runs with paleoclimate to those with fixed
climate we calculate that carbon sequestration due to
climate reached up to 3 Pg during the past 500 years
(Fig. 2). During the four centuries before 1900, temperatures in Europe were generally below the 20th century
average, with minima centered around 1700 and the
late 19th century (Luterbacher et al., 2004). During these
cold periods carbon sequestration due to climate was
greatest (Fig. 2). Cooler temperatures during the LIA
suppressed primary productivity and led to modest
Table 1 Simulated carbon in living biomass, soil and litter, and total carbon in terrestrial ecosystems of Europe at 1500, 1850, 1950,
and 2000 in LPJ-DGVM runs with fixed climate and reconstructed paleoclimate. Results are shown for the natural vegetation (control) run and for each of the ALCC scenarios implemented in this study. All runs were performed with transient CO2 concentrations
and all values are in units of Pg C
Climate scenario
Living biomass C
Year
1500
1500
1850
1950
2000
1500
1850
1950
2000
Percent
change
Fixed
Paleo
Natural vegetation (no land use)
50.6
50.5
53.2
57.4
166.5
50.5
50.0
52.6
55.7
166.6
168.6
170.1
172.2
173.0
175.2
174.3
217.0
217.1
219.2
220.0
225.4
225.6
232.6
230.0
6.7
5.6
Fixed
Paleo
KK10 land use merged to HYDE (Kaplan et al., 2009; Klein-Goldewijk et al., 2010)
33.4
29.7
30.7
34.9
143.6
139.7
139.0
140.7
177.0
169.3
33.4
29.0
29.9
33.3
143.6
140.8
139.6
140.0
177.0
169.8
169.7
169.5
175.6
173.4
0.8
2.1
Fixed
Paleo
KK10 land use interpolated to HYDE (Kaplan et al., 2009; Klein-Goldewijk et al., 2010)
33.4
30.3
32.9
36.4
143.5
140.3
142.5
145.4
176.9
170.6
175.4
33.4
29.7
32.1
34.9
143.6
141.4
143.1
144.8
177.0
171.1
175.2
181.9
179.7
2.7
1.5
Fixed
Paleo
HYDE land use (Klein-Goldewijk et al., 2010)
44.2
37.1
31.4
34.6
157.9
151.2
44.1
36.4
30.7
33.0
158.0
152.4
176.7
174.5
14.3
15.8
1850
1950
Soil and litter C
2000
© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
Total C
143.3
143.9
142.2
141.5
202.1
202.1
188.2
188.8
174.7
174.6
Annual temperature anomaly (°C)
906 K A P L A N e t a l .
1.0
effect, however, is smaller for runs with simulated
ALCC, as there is less potential for cropland ecosystems to absorb carbon than for forests and annual harvest does not allow accumulation of carbon in the
aboveground biomass and litter. These runs with
ALCC forcing indicate that CO2 fertilization could have
led to the sequestration of up to 12 Pg of carbon in
European ecosystems (Fig. 3; Table S1).
(a)
0.5
0.0
−0.5
ALCC effects on the terrestrial biosphere
−1.0
−1.5
Natural vegetation (no land use)
KK10 land use, merged to HYDE
KK10 land use, interpolated to HYDE
HYDE land use
1
Δ C storage (Pg)
0
−1
−2
−3
(b)
−4
1500
1600
1700
1800
1900
2000
Year
Fig. 2 Climate effects on carbon storage. (a) European mean
annual temperature anomaly from the datasets used to drive
LPJ, the heavy line shows temperature smoothed with a 20 year
Gaussian filter to highlight decadal trends, and (b) simulated
changes in terrestrial carbon storage due to climate variability;
negative values indicate a sink. Net emissions are calculated as
the difference in carbon storage between model runs with fixed
climate and reconstructed paleoclimate.
reductions in carbon stored in living biomass. However, microbial decomposition of litter and soil organic
matter was simultaneously suppressed; this reduction
in heterotrophic respiration generally led to increasing
soil carbon storage and a positive net flux of carbon into
the biosphere (Fig. S2; Table 1).
Increases in atmospheric CO2 concentration from
preindustrial (approximately 280 ppm) to late 20th century levels (approximately 370 ppm) had a larger effect
on carbon storage than climate. Results of experiments
we performed with natural vegetation where climate
was held under constant, mean 20th century conditions, and comparisons between model runs with fixed
and transient CO2 clearly show a strong CO2 fertilization after 1800. CO2 increases since the mid-19th century could have added up to approximately 15 Pg to
the carbon inventory of Europe (Fig. 3; Table 1). This
The impact of ALCC on the terrestrial carbon balance
of Europe was greater than that of both climate and
CO2 over the past 500 years. All ALCC scenarios
result in carbon storage well below potential natural
values at 1500, followed by further losses of terrestrial
carbon stored in European ecosystems during the first
part of the simulation (Fig. 3; Table 1). Subsequently,
all scenarios also show a reversal from carbon loss to
accumulation caused by a combination of land abandonment and CO2 fertilization. Relative to year 1500,
the effects of ALCC alone could have ranged from
5 Pg of carbon uptake to 25 Pg of carbon emissions
over the last 500 years, depending on the ALCC scenario (Fig. 3; Table 1). This large range is mostly due
to the differing amounts of ALCC between the HYDE
and KK10 scenarios at 1500.
Preindustrial land use estimates in KK10 contrast
starkly with those depicted in the HYDE scenario (Kaplan et al., 2010; Fig. 1). Most ALCC in the KK10 scenarios occurs prior to the 16th century and therefore a
highly deforested landscape was depicted at the beginning of our runs at 1500. Specifically, the biosphere contained 40 Pg less C than the potential natural state
when using the KK10 scenarios (Fig. 3; Table 1). LPJ
was spun up to equilibrium with year 1500 ALCC estimates and it is likely that European croplands were
nearly in a steady state with respect to carbon since,
according to the KK10 scenario, most areas were deforested during prehistoric times and the carbon in soils
presumably had enough time to equilibrate after conversion from forest to cropland.
In contrast to the time progression of ALCC in KK10,
the HYDE scenario runs begin at 1500 with a carbon
inventory only 15 Pg C below potential natural levels.
Although HYDE was spun up to equilibrium at year
1500 ALCC estimates, soils could have contained additional carbon from forests that were cut just prior to
1500, when deforestation in Europe, according the
HYDE scenario, was just beginning to accelerate (Klein
Goldewijk et al., 2011). Nevertheless, carbon losses due
to ALCC after 1500 are rapid for simulations with the
HYDE ALCC relative to those with the KK10 scenario
(Fig. 3).
© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
T H E E F F E C T S O F L A N D U S E A N D C L I M A T E C H A N G E 907
Natural vegetation
KK10−merged land use
KK10−interpolated land use
HYDE land use
230
220
220
210
210
200
200
190
190
180
180
170
170
(a)
Terrestrial C storage (Pg)
Terrestrial C storage (Pg)
230
(b)
160
160
1500
1600
1700
1800
1900
2000 1500
Year
1600
1700
1800
1900
2000
Year
Fig. 3 Timeseries of carbon storage for runs with transient CO2 (a) and fixed CO2 (b). Each line represents simulated total carbon in
vegetation and soils under different land use scenarios over the time period 1500 to 2000. Heavy lines represent runs with paleoclimate;
model runs with fixed climate are shown with thin lines. The difference in carbon storage at year 1500 reflects the land use map used to
initialize the simulation (see Fig. 1).
Combined effects of climate variability, increases in
atmospheric CO2, and ALCC
When climate change, transient CO2 concentrations,
and ALCC were combined as input into LPJ, we
observe that terrestrial European carbon stocks at year
2000 are 55 Pg below what they would potentially be
under natural vegetation (Fig. 3); approximately
175 Pg compared to approximately 230 Pg, respectively. In combination with ALCC, climate change
caused one-third less carbon to be sequestered during
cold periods compared to runs with natural vegetation
(Fig. 2). CO2 fertilization integrated with ALCC and
climate change caused nearly 5 Pg of additional carbon
to be sequestered in terrestrial ecosystems by 1950
with an additional 7 Pg of carbon by 2000 (Tables 1
and S1).
Runs with the KK10 land use scenarios reach a minimum in C storage at roughly 170 Pg, whereas the low
point in the HYDE simulations occurs later and with a
slightly larger terrestrial carbon stock of 175 Pg
(Fig. 3). Following these nadirs in terrestrial biomass, a
reversal to carbon accumulation occurs, albeit at different times for each of our three ALCC scenarios. The
two types of extensions (merged vs. interpolated) used to
bring the KK10 scenarios to year 2000 resulted in notable differences for the carbon balance at year 2000. For
the KK10-merged scenario, emissions from ALCC are
© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
fairly stable from 1800 until the mid-20th century, at
which point a reversal to carbon accumulation occurs.
As the last 40 years of the KK10-merged dataset contains the same land use estimates as HYDE, the HYDE
scenarios show the same reversal point, at around
1960.
Carbon fluxes simulated in the combined simulations were highly variable (up to 50 Tg amplitude)
year-on-year; this amount is consistent with other
studies. During the last 50 years of our simulations,
simulated net biome production (NBP) ranged from 50
–450 TgC yr 1. A recent modeling study by Churkina
et al. (2010) accessed the effects of land cover conversion, nitrogen deposition, and climate change on the
carbon balance of Europe, concluding that European
land ecosystems absorbed 56 Tg of carbon per year
on average during the period 1950–2000 and
114 Tg C yr 1 for the early 21st century. Other databased studies estimated that European rates of C
sequestration in terrestrial ecosystems in recent decades
range from 111 Tg C yr 1 (Janssens et al., 2003) to
313 Tg C yr 1 (Schulze et al., 2009).
Spatial variability in European carbon stocks
Maps of changing carbon storage aid to illustrate the
effects of ALCC over the past 500 years (Fig. 4). All
scenarios show a general shift in the locus of anthropo-
908 K A P L A N e t a l .
20th century
18th & 19th centuries
HYDE
KK10, interpolated to HYDE
KK10, merged to HYDE
16th & 17th centuries
−5
−4
−3
−2
−1
0
1
2
3
4
5
Terrestrial carbon gain or loss over time period (kg m−2)
Fig. 4 Maps of carbon dynamics. Difference in carbon inventory between the beginning and the end of a time period (1501–1700, 1701–
1900, 1901–2000) for the KK10-merged, KK10-interpolated and HYDE land use scenarios. Positive values indicate carbon emissions. All
simulations were run with paleoclimate forcing.
genic land use from west to east, but differ in the pattern and timing of carbon emissions and uptake. In the
KK10 scenarios, the 16th and 17th centuries are characterized by a heterogeneous pattern with both highest
emissions and uptake (±5 kg C m 2) in montane areas
of central Europe, partly because the reconstructed climate anomalies are greatest in mountain areas
(Luterbacher et al., 2004; Pauling et al., 2006). From the
18th century onwards, carbon emissions shift eastward
as increasing population densities and new technologies opened up the steppe and forest steppe regions of
Eastern Europe to cultivation. Moderate levels of carbon uptake (2 kg C m 2) occur during this time period
in western and central Europe in all ALCC scenarios,
with nearly zero emissions occurring in this region
when using the KK10-interpolated scenario (Fig. 4, middle panels). By the 20th century the carbon-sequestering effects of land abandonment and CO2 fertilization
are apparent in the KK10-interpolated scenario across
nearly all of Europe (1–5 kg m 2 of uptake), whereas
the HYDE scenario still shows widespread conversion
to anthropogenic land use and high carbon emissions
(1–5 kg m 2 of emissions) in the east. The KK10-merged
scenario simulation depicts mostly uptake during the
20th century with only minor emissions occurring in
the mountainous areas of Europe and Ireland.
© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
T H E E F F E C T S O F L A N D U S E A N D C L I M A T E C H A N G E 909
Discussion
Climate and CO2 effects on carbon
Climate, CO2, and ALCC had contrasting yet sometimes
synergistic effects on the European carbon budget over
the last 500 years. The effect of climate variability alone
was dampened when using an ALCC scenario because
large parts of Europe that would naturally be dominated by carbon-rich temperate forests were replaced
by agricultural land, which contains less biomass and
soil carbon (Fig. 2b). The effect of increasing atmospheric CO2 concentrations had a strong effect on productivity and carbon storage in our simulations,
reaching over 15 Pg in the natural vegetation experiment (Fig. 3; Tables 1 and S1). It is possible that this
CO2 fertilization effect could be overestimated, as nitrogen availability can limit carbon sequestration caused
by increasing CO2 (Fisher et al., 2010). On the other
hand, anthropogenic N deposition has been widespread
and particularly strong in Europe since industrialization (Galloway et al., 2004), coincident with the timing
of increasing CO2, and there is empirical evidence that
increasing atmospheric CO2 concentrations over the
past 150 years did positively affect the growth of forest
trees (Gedalof & Berg, 2010). Thus, much of the C-fertilization potential simulated by our model may have
been or yet be realized (Magnani et al., 2007; Zaehle
et al., 2010). We would not necessarily expect this affect
to continue into the future, because although CO2 and
N deposition continues to rise, N loading has become
critical in some regions (e.g., Schulze et al., 1989), and
other mineral nutrients will also limit growth.
ALCC effects on carbon
The time trend of land use in the different ALCC scenarios has the strongest effect on carbon storage among
all of the factors we tested. The timing of the reversal
from deforestation to afforestation is particularly
important for the current state of European ecosystems.
That carbon storage in the HYDE and KK10-merged simulations reaches a low point in the mid-20th century is
an artifact of the methodology used to produce the
HYDE dataset, where FAO statistics on crop and pasture areas only extend back to 1961 (FAO, 2008; Klein
Goldewijk et al., 2011). Conversely, the KK10-interpolated scenario shows a prominent reversal to carbon
accumulation at roughly 1850 – see also our discussion
below on forest transitions – which is a result of the linear interpolation from KK10 1850 land use to year 2000
land use in HYDE. It is important to note that without
CO2 fertilization, no reversal to carbon accumulation
occurs for any of the ALCC scenarios. However, with© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
out land abandonment, carbon storage due to CO2 fertilization would have been less important.
Although the ALCC scenario implemented strongly
affects the simulation of preindustrial carbon balances,
surprisingly, the resulting carbon inventory of Europe
at year 2000 does not differ greatly between ALCC
datasets used in this study (Fig. 3). The KK10-interpolated scenario makes the central assumption that European countries experienced forest transitions during
the mid-19th century (Mather et al., 1998a; Mather,
1999; Mather & Fairbairn, 2000). This results in greater
carbon inventories in Europe 150 years later, at approximately 180 Pg, roughly 5 Pg more than simulations
with our other two ALCC scenarios. As a result of identical 1961–2000 land use estimates, the small differences
in carbon balances at year 2000 between KK10-merged
scenario and HYDE are due to ALCC prior to 1961
(Fig. 3). For example, fixed climate simulations with
the HYDE scenario result in 1.1 Pg less terrestrial carbon at year 2000 than the KK10-merged scenario (Fig. 3;
Table 1). This occurs because according to the HYDE
scenario, lands were deforested more recently and the
soil on previously forested land still contains and continues to emit carbon even at year 2000 due to the slow
decomposition of soil organic matter. The large discrepancy among ALCC estimates for Europe even over the
last 150 years (Table 2) is a major source of uncertainty
in our results, as was also shown in previous studies (e.
g., Zaehle et al., 2007; Churkina et al., 2010).
Even so, the narrow range in simulated carbon storage at year 2000 in our results indicates that ALCC
occurring over the past one to two centuries following
industrialization had the strongest influence on the
present state of the biosphere, outweighing most effects
of preindustrial ALCC. In our model simulations, the
mean residence time of carbon in European terrestrial
ecosystems is rather short, on the order of 200 years.
As soil organic matter can be stabilized in clay minerals (Six et al., 2002), the amount of soil organic matter
entering the slow soil turnover (1000-year) pool may be
strongly influenced by the total clay content of the soil
(Jobbágy & Jackson, 2000). The relatively young soils of
much of Europe north of the Alps have relatively low
clay contents. Thus, in the version of LPJ we used, carbon storage in Europe is predominantly in the living
biomass and the fast-turnover SOM pool, both of which
have mean residence times on the order of a few centuries. On the other hand, synthesis of the mean age of
soil organic matter based on radiocarbon dating of bulk
soils (Fig. S3) indicates that much older soil organic
matter can be observed in European ecosystems. It
could be that our version of LPJ, while producing a
geographic distribution of soil organic matter stocks
that is much more in line with observations than previ-
910 K A P L A N e t a l .
Table 2 A compilation of published agricultural areas (106 ha) in Europe since 1850. N.B. the exact geographic definition of Europe varies slightly between studies, but always excludes the former Soviet Union
Area considered
Year
1850
1870
1900
1910
1950
1990
2000
Robertson
(1956):
arable
area
509
140
145
148
Houghton
(1999):
cropland
CORINE land
cover: Arable
land and
permanent
crops
CORINE land
cover:
Pastures
and mosiacs
KK10merged:
land use
HYDE:
crop +
pasture
Ramankutty
et al. (2008):
Cropland
Ramankutty
et al. (2008):
Pasture
430
373
373
481
492
467
467
82
82
272
265
259
257
239
225
214
180
190
209
217
232
227
216
125
67
61
70
74
76
82
73
117
117
ous versions of this model, requires further modification, such as the inclusion of a third soil organic matter
pool (e.g., Baisden & Amundson, 2003) to better simulate the temporal dynamics of carbon turnover.
We further explored the assumption we made
regarding the equilibrium state of the biosphere in Europe at 1500 CE by performing a series of additional
fixed-climate simulations started at 1000 BC (3 ka) and
comparing European carbon storage simulated at year
1500 in a transient run with that of our standard model
runs, where the model was spun-up to equilibrium at
1500 CE (Fig. S4). These experiments show that in most
of Europe, especially when using the KK10 scenario,
carbon storage was close to equilibrium conditions at
1500 despite a long history of land conversion and
abandonment in the preceding millennia. As discussed
above, the residence time of carbon in most European
ecosystems is simulated to be ca. 200 years on average,
so only ALCC in the preceding centuries will have a
substantial impact on carbon storage at 1500. Where
substantial dis-equilibrium is apparent (Fig. S4), it was
caused by rapid land cover change during the two to
three centuries preceding 1500. This dis-equilibrium
appears in the southern Balkans and Russia in the
KK10 scenario and in much of Western Europe in the
HYDE scenario. In these regions, rapid deforestation in
the period 1200–1350 CE was followed by widespread
abandonment after the Black Death (1350–1400) and
renewed expansion of agricultural areas before 1500.
The magnitude of the pre-1500 fluctuations in land use
is particularly strong in the HYDE scenario.
Combined effects
When climate change, transient CO2 concentrations,
and ALCC were combined as input into LPJ it becomes
evident that the effects of ALCC dominate the terrestrial European carbon cycle over the past 500 years.
Climate variability, both cold periods during the LIA
and the late 20th century warming, caused short-term
fluctuations in total ecosystem carbon, with colder periods leading to increased storage, and vice versa. As discussed above, the relatively short residence time of
carbon in most European terrestrial ecosystems means
that the LIA-driven increases in carbon storage do not
persist into the latter half the 20th century. CO2 fertilization over the 20th century led to substantial
increases in carbon stocks and offset some of the carbon
lost due to both warming and ALCC. The trends in carbon storage described in the previous section on ALCC
effects on carbon storage are still the predominant control on European carbon storage even in model experiments driven by variable climate and CO2.
For model runs using the KK10-interpolated ALCC
scenario, all carbon emissions are offset by land abandonment and CO2 fertilization and the resulting year
2000 European carbon stock is 1.5% higher than in 1500
(Table 1). However, using fixed CO2 concentrations of
the 16th century, the carbon balance for this simulation
is still 5.4% below the 1500 carbon stock. The HYDE
and KK10-merged scenarios both resulted in a net loss of
carbon over the 500-year time period, but still show a
reversal to carbon accumulation after 1950. Interestingly, by holding CO2 levels at preindustrial levels, the
reversal to carbon accumulation becomes nonexistent
under all ALCC scenarios, demonstrating the importance that CO2 fertilization has on the carbon balance of
Europe at year 2000 (Tables 1 and S1). Churkina et al.
(2010) found that models that account for carbonnitrogen dynamics showed a smaller response to CO2
fertilization, attributed to nutrient limitation. Including
nitrogen dynamics in LPJ could result in a weaker
© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
T H E E F F E C T S O F L A N D U S E A N D C L I M A T E C H A N G E 911
response to increases in atmospheric CO2 over the last
150 years, and therefore lead to a diminished response
of C storage to land abandonment during the last century of our model simulations.
The effects of climate variability are low for simulations with natural vegetation, but even lower when
ALCC is simulated as well. Even so, cold periods
during the LIA appear to have had a synergistic
effect with ALCC prior to 1950 by allowing for temporary sequestration of additional carbon in the form
of soil organic matter. However, this carbon is rapidly emitted when the climate warms during the 20th
century (Table 1), leaving no long-lasting effect (see
natural vegetation runs in Fig. S2; Table 1). According
to Zaehle et al. (2007), further carbon emissions from
soil respiration are likely to continue with increased
warming over the 21st century. After 1950, in addition to increased carbon emissions from soils, biomass
production also decreased compared to fixed climate
runs likely due to increased evapotranspiration from
warmer temperatures without significant increases in
precipitation during summer (Pauling et al., 2006;
Table 1). This adversely affects any carbon sequestration from late 20th century land abandonment occurring in the HYDE and KK10-merged scenarios. Overall,
the effect of climate change combined with ALCC is
very small in our experiments, less than 1 Pg on
average (Fig. 3, difference between thin and thick
lines).
Forest transitions
By comparing the outcomes of model runs using our
different ALCC scenarios, we can speculate on the
importance of the timing of the forest transition. The
HYDE dataset depicts sharply increasing ALCC until
ca. 1960 when a sharp transition to land abandonment
occurs (Fig. 1d). The KK10-interpolated dataset has a visible transition nearly 100 years earlier, an intentional
artifact of a straight linear interpolation to year 2000
land use fractions, implying a ubiquitous European forest transition at 1850. The KK10-merged dataset is essentially a combination of the HYDE and KK10-interpolated
datasets, as KK10 data are blended to HYDE ALCC
fractions. This dataset shows three reversals from
deforestation to afforestation, at year 1850, 1900, and
another one at 1960, as in HYDE. The reversal points in
all three datasets inherently have a strong influence on
the outcome of our LPJ simulations and have a substantial effect on the year 2000 carbon balance. However, a
breakdown of ALCC on a per country basis reveals that
neither the HYDE nor the KK10 ALCC datasets capture
the forest transition in most countries as it was
recorded in contemporary observations (Fig. S5).
© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
Despite numerous studies performed on forest transition phenomenon (Mather, 1992, 1999; Mather &
Needle, 2000; Rudel et al., 2005; Kauppi et al., 2006), the
timing and magnitude of the forest transition is not
adequately represented in any of the ALCC datasets we
used (Fig. S5). Furthermore, land use inventories disagree over the amount of agricultural land area in Europe for the last two centuries (Table 2), and all current
land use models underestimate the amount of ALCC at
the time of the transition for all regions with a documented forest transition (Fig. S5, triangles). Forest transitions in western and Central Europe occurred on
average around 1861 (area-weighted average of documented cases for France, Denmark, Switzerland, Ireland, the UK, Sweden, Portugal, and Italy). In Eastern
Europe an area-weighted average of Hungary and
European Russia puts the average timing of the forest
transition at 1930. From these years onward, we expect
to see increasing carbon inventories, as agricultural
lands are abandoned and ALCC decreases. If the magnitude of ALCC and the timing of the forest transitions
were captured correctly, as according to the historical
records, then the results of our simulations would
likely be significantly different, possibly showing substantially more carbon accumulation in European ecosystems over the past 150 years.
The present and future carbon balance of Europe
Our natural vegetation scenario demonstrates the
potential amount of carbon that could be stored in Europe in the unlikely scenario that all land is abandoned
and left to regrow with natural vegetation. We found
it more informative to examine only those lands in
Europe under natural vegetation in year 2000 (i.e., not
currently used for crops or pasture). We calculated the
difference between the carbon inventories at year 2000
for these unused areas between the natural vegetation
scenario and the ALCC scenarios. This provides an
indication of the carbon sequestration potential of current European natural ecosystems in future decades.
Considering only the area of Europe that is currently
not under ALCC and assuming no future changes in
land use, we estimate that between 5 and 12 Pg of carbon remain to be sequestered. As a result of the relatively early forest transition in the KK10-interpolated
scenario, at present, forest ecosystems are closer to saturation with respect to carbon (cf. Albani et al., 2006).
In contrast, model runs using the HYDE and KK10merged scenario, where deforestation peaked only in
the second half of the 20th century indicate that terrestrial ecosystems are still far from equilibrium and
therefore have more potential to absorb carbon. These
results are in line with the results of future scenario
912 K A P L A N e t a l .
model simulations that indicated similar amounts of
sequestration (Zaehle et al., 2007).
To further clarify the effects of past land use on European carbon stocks at the present day, we compared a
LPJ run spun up to equilibrium with year 2000 land use
and CO2 concentrations with a run spun up to year
1500 and run transiently with the KK10-merged scenario
and transient CO2 concentrations, 1500–2000. The difference between these two runs demonstrates the effect
of land use on carbon at year 2000 (Fig. 5). ALCC prior
to 2000 caused less carbon to be stored in Ireland,
northern European Russia, and in smaller regions of
western and central Europe (Fig. 5; areas in red). These
areas are depleted with respect to equilibrium carbon
storage of up to approximately 10 kg m 2 C, and therefore would have the potential to sequester additional
carbon in the future. In contrast, green areas on Fig. 5
highlight areas where more carbon is stored in ecosystems according to the LPJ run that included ALCC. In
this case, soils contain more carbon from previously
forested areas; this is carbon that was not stored in the
run spun up with year 2000 land use patterns. This carbon would continue to be decomposed and emitted
after year 2000. Taken in sum, the whole of our European study region has the potential to take up an addi-
KK10 merged at 2000
− spinup to 2000
tional 12 Pg C (Fig. 5), not accounting for future ALCC
or CO2 fertilization.
Previous studies support the claim that European
ecosystems have been a net sink of carbon over the past
several decades and will continue to be in the near
future (Vermoere et al., 2000; Janssens et al., 2003;
Churkina et al., 2010). Ciais et al. (2008) published a
study on carbon accumulation in European forests in
which forest inventories and carbon accumulation rates
from the EU 15 countries (plus Switzerland and Norway, excluding Luxembourg) were accessed over the
past 50 years. They report that biomass carbon stocks
and NPP in European forests increased 1.75-fold and
1.67-fold, respectively, from 1950 to 2000. Over the
same time period, in our model simulations we demonstrate roughly a 1.3-fold increase in carbon stocks and a
1.2-fold increase in NPP. Differences between Ciais
et al. (2008) and our study can be attributed to (1)
uncertainties in historical ALCC data (see previous section), (2) the comparatively coarse resolution used in
this study (i.e., a single 30 min gridcell is not typically
entirely forest, even under natural conditions, and
therefore average biomass is typically less than
reported for just forests), (3) nonforested areas included
in our study (e.g., Mediterranean biomes, Eastern European steppe) and (4) lack of forest management in our
simulations, as the net biomass sink reported in Ciais
et al. (2008) is largely attributed to forestry and harvesting rates, rather than increases in forest area. Our
study, conversely, focuses on the overall carbon stocks
in European ecosystems, rather than just forests, and a
substantial reason for the European carbon sink can be
attributed to a combination of CO2 fertilization plus
increases in the area of abandoned agricultural land.
Conclusions and future directions
−10
−8
−6
−4
−2
0
2
4
6
8
10
Carbon difference (kg m−2)
Fig. 5 Terrestrial carbon lost or gained as a result of the past
500 years of ALCC. A LPJ run spun up to equilibrium at year
2000 conditions (with respect to ALCC and CO2 concentration)
was subtracted from year 2000 of a 500 year run with transient
CO2 and KK10-merged land use. Areas in red have less carbon,
while areas in green have more carbon as a result of anthropogenic land use.
In this study, we demonstrate that ALCC clearly overshadows the effect of climate variability on the European carbon cycle over the past 500 years. After a long
history of net carbon loss to the atmosphere through
deforestation, Europe experienced gradual abandonment of marginal lands and afforestation in the 19th
and 20th centuries. Increasing atmospheric CO2 concentrations, supported by forest regrowth, and in some
areas, warmer climate, resulted in the reversal of some
or all of the carbon lost since 1500. The timing of this
reversal, as reflected in the different ALCC scenarios, is
especially important for assessing the trajectory of ecosystems at present and hence their remaining sequestration potential, although all of the ALCC scenarios
result in simulations that indicate European forests still
have the potential to sequester atmospheric carbon in
the future.
© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
T H E E F F E C T S O F L A N D U S E A N D C L I M A T E C H A N G E 913
There are several limitations to our current study that
should be addressed in future work on assessing the
state of the European carbon cycle. These include adding the capability to simulate forest management, cultivation practices such as tilling, nitrogen deposition,
fertilization, and nitrogen cycle dynamics to LPJ. In
addition, it is absolutely necessary to improve ALCC
datasets to better reflect the forest transitions and
cleared land amounts recorded in the historical record
during the last two centuries.
Our current inability to quantify the effects of industrial forest management on the carbon inventory on
vegetation and soils, e.g., as a result of the large-scale
logging and draining of peat forests that was widespread in Northern Europe over the 20th century (e.g.,
Jandl et al., 2007) could have an important affect on our
final results. Future studies could build on vegetation
models that contain schemes for simulating forest management (Zaehle et al., 2007). Our calculations of
sequestration potential also do not take into account the
potential for enhanced carbon sequestration as a
result of future CO2 fertilization, although this effect
may be limited by lack of nitrogen or other nutrients
(Thornton et al., 2007; Gedalof & Berg, 2010) and from
tropospheric ozone damage to plants (Sitch et al.,
2007). Nevertheless, this study demonstrates the
importance of accounting for at least centennial-scale
changes in land use, CO2 and climate in evaluating
the current state and potential future behavior of
European ecosystems.
Acknowledgements
This work was supported by a Swiss National Science Foundation professorship grant (PP0022_119049) to JOK, FIRB project
CASTANEA (RBID08LNFJ), the EU MILLENNIUM IP (GOCE017008), and CCES project MAIOLICA. We thank Helga Vanthournout for her comments that improved this manuscript,
and Sue Trumbore, Troy Baisden, Jan Esper, and the EPFL
ARVE Group for discussions.
Churkina G, Zaehle S, Hughes J et al. (2010) Interactions between nitrogen deposition,
land cover conversion, and climate change determine the contemporary carbon
balance of Europe. Biogeosciences, 7, 2749–2764.
Ciais P, Schelhaas MJ, Zaehle S et al. (2008) Carbon accumulation in European forests.
Nature Geoscience, 1, 425–429.
Denman KL, Brasseur G, Chidthaisong A et al. (2007) Couplings between changes in
the climate system and biogeochemistry. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (eds Solomon S, Qin D, Manning M, Chen Z,
Marquis M, Averyt KB, Tignor M, Miller HL), pp. 499–587. Cambridge University
Press, Cambridge, New York.
Ebermayer E (1876) Die Gesammte Lehre der Waldstreu mit Rücksicht auf die Chemische
Statik des Waldbaues. Julius Springer, Berlin.
Ellenberg H (1988) Vegetation Ecology of Central Europe. Cambridge University Press,
Cambridge.
FAO (2008) FAOSTAT, Food and Agriculture Organization of the United Nations (FAO).
FAO, Rome, Italy, Available at: http://faostat.fao.org (accessed 31 October 2011).
FAO/IIASA/ISRIC/ISSCAS/JRC (2009) Harmonized World Soil Database (version 1.1).
FAO, Rome, Italy and IIASA, Laxenburg, Austria.
Fisher JB, Sitch S, Malhi Y, Fisher RA, Huntingford C, Tan S-Y (2010) Carbon cost of
plant nitrogen deposition: a mechanistic, globally applicable model of plant nitrogen uptake, retranslocation, and fixation. Global Biogeochemical Cycles, 24, GB1014.
Friedlingstein P, Cox P, Betts R et al. (2006) Climate-carbon cycle feedback analysis:
results from the C4MIP model intercomparison. Journal of Climate, 19, 3337–3353.
Galloway JN, Dentener FJ, Capone DG et al. (2004) Nitrogen cycles: past, present, and
future. Biogeochemistry, 70, 153–226.
Gedalof Z, Berg AA (2010) Tree ring evidence of limited direct CO2 fertilization of
forests over the 20th century. Global Biogeochemical Cycles, 24, GB3027, doi:
10.1029/2009GB003699.
Glatzel G (1991) The impact of historic land use and modern forestry on nutrient relations of Central European forest ecosystems. Fertilizer Research, 27, 1–8.
Gurney KR, Law RM, Denning AS et al. (2004) Transcom 3 inversion intercomparison: model mean results for the estimation of seasonal carbon sources and sinks.
Global Biogeochemical Cycles, 18, GB1010.
Houghton RA (1999) The annual net flux of carbon to the atmosphere from changes
in land use 1850-1990. Tellus, 51B, 298–313.
Jain AK, Yang X (2005) Modeling the effects of two different land cover change data
sets on the carbon stocks of plants and soils in concert with CO2 and climate
change. Global Biogeochemical Cycles, 19, GB2015, doi: 10.1029/2004GB002349.
Jandl R, Lindner M, Vesterdal L et al. (2007) How strongly can forest management
influence soil carbon sequestration? Geoderma, 137, 253–268.
Janssens IA, Freibauer A, Ciais P et al. (2003) Europe’s terrestrial biosphere absorbs 7
to 12% of European anthropogenic CO2 emissions. Science, 300, 1538–1542.
Jobbágy EG, Jackson RB (2000) The vertical distribution of soil organic carbon and its
relation to climate and vegetation. Ecological Applications, 10, 423–436.
Jungclaus JH, Lorenz SJ, Timmreck C et al. (2010) Climate and carbon-cycle variability over the last millennium. Climate of the Past, 6, 723–737.
Kaplan JO, Krumhardt KM, Zimmerman N (2009) The prehistoric and preindustrial
deforestation of Europe. Quaternary Science Reviews, 28, 3016–3034.
Kaplan JO, Krumhardt KM, Ellis EC, Ruddiman WF, Lemmen C, Goldewijk KK
(2010) Holocene carbon emissions as a result of anthropogenic land cover change.
The Holocene, 21, 775–791, doi: 10.1177/0959683610386983.
Kauppi PE, Ausubel JH, Fang J, Mather AS, Sedjo RA, Waggoner PE (2006) Returning
forests analyzed with the forest identity. Proceedings of the National Academy of
Albani M, Medvigy D, Hurtt GC, Moorcroft PR (2006) The contributions of land-use
change, CO2 fertilization, and climate variability of the Eastern US carbon sink.
Global Change Biology, 12, 2370–2390.
Baisden WT, Amundson R (2003) An analytical approach to ecosystem biogeochemistry modeling. Ecological Applications, 13, 649–663.
Sciences, 103, 17574–17579.
Klein Goldewijk K (2001) Estimating global land use change over the past 300 years:
the HYDE Database. Global Biogeochemical Cycles, 15, 417–433.
Klein Goldewijk K, Beusen A, De Vos M, Van Drecht G (2011) The HYDE 3.1 spatially
explicit database of human induced land use change over the past 12,000 years.
Global Ecology & Biogeography, 20, 73–86.
Kozak J, Estreguil C, Troll M (2007) Forest cover changes in the northern Carpathians
Batjes NH (2006) ISRIC-WISE derived soil properties on a 5 by 5 arc-minutes global
grid. Report 2006/02, ISRIC- World Soil Information, Wageningen, Available at:
http://www.isric.org (accessed 31 October 2011).
Bradshaw R (2008) Detecting human impact in the pollen record using data-model
comparison. Vegetation History and Archaeobotany, 17, 597–603.
Büntgen U, Tegel W, Nicolussi K et al. (2011) 2500 Years of European climate variabil-
in the 20th century: a slow transition. Journal of Land Use Science, 2, 127–146.
Krumhardt KM, Kaplan JO (2010) ARVE Technical Report #2: A spline fit to atmospheric CO2 records from Antarctic ice cores and measured concentrations for the
past 25,000 years. Environmental Engineering Institute, Ecole Polytechnique
Fédérale de Lausanne, Lausanne, Switzerland.
Luterbacher J, Dietrich D, Xoplaki E, Grosjean M, Wanner H (2004) European sea-
ity and human susceptibility. Science, 331, 578–582.
Canadell JG, Pataki DE, Gifford R et al. (2007) Saturation of the terrestrial carbon sink.
In: Terrestrial Ecosystems in a Changing World (eds Canadell JG, Pataki DE, Pitelka
LF), pp. 59–78. Springer, Heidelberg.
sonal and annual temperature variability, trends and extremes since 1500. Science,
303, 1499–1503.
Luyssaert S, Schulze ED, Borner A et al. (2008) Old-growth forests as global carbon
sinks. Nature, 455, 213–215.
References
© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914
914 K A P L A N e t a l .
Magnani F, Mencuccini M, Borghetti M et al. (2007) The human footprint in the carbon cycle of temperate and boreal forests. Nature, 447, 848–850.
Strassmann KM, Joos F, Fischer G (2008) Simulating effects of land use changes on
carbon fluxes: past contributions to atmospheric CO2 increases and future commit-
Mather AS (1992) The forest transition. Area, 24, 367–379.
Mather AS (1999) The course and drivers of the forest transition: the case of France.
Journal of Rural Studies, 15, 65–90.
Mather AS, Fairbairn J (2000) From floods to reforestation: the forest transition in
Switzerland. Environment and History, 6, 399–421.
Mather AS, Needle CL (2000) The relationships of population and forest trends. The
Geographical Journal, 166, 2–13.
ments due to losses of terrestrial sink capacity. Tellus, 60B, 583–603.
Thornton PE, Lamarque JF, Rosenbloom NA, Mahowald NM (2007) Influence of
carbon-nitrogen cycle coupling on land model response to CO2 fertilization and
climate variability. Global Biogeochemical Cycles, 21, GB4018, doi: 10.1029/
2006GB002868.
Trumbore S (2000) Age of soil organic matter and soil respiration: radiocarbon constraints on belowground C dynamics. Belowground Processes and Global Change, 10,
Mather AS, Needle CL, Coull JR (1998a) From resource crisis to sustainability: the forest transition in Denmark. International Journal of Sustainable Development, 5, 182–
193.
Mather AS, Needle CL, Fairbairn J (1998b) The human drivers of global land cover
change: the case of forests. Hydrological Processes, 12, 1983–1994.
McGuire AD, Sitch S, Clein JS et al. (2001) Carbon balance of the terrestrial
399–411.
Vermoere M, Smets E, Waelkens M, Vanhaverbeke H, Librecht I, Paulissen E, Vanhecke L (2000) Late Holocene environmental change and the record of human
impact at Gravgaz near Sagalassos, southwest Turkey. Journal of Archaeological Science, 27, 571–595.
Zaehle S, Bondeau A, Carter TR et al. (2007) Projected changes in terrestrial carbon
biosphere in the twentieth century: analyses of CO2, climate and land use effects
with four process-based ecosystem models. Global Biogeochemical Cycles, 15, 183–
206.
Mitchell TD, Jones PD (2005) An improved method of constructing a database of
monthly climate observations and associated high-resolution grids. International
Journal of Climatology, 25, 693–712.
Olofsson J, Hickler T (2008) Effects of human land-use on the global carbon cycle dur-
storage in Europe under climate and land-use change, 1990-2100. Ecosystems, 10,
380–401.
Zaehle S, Friend AD, Friedlingstein P, Dentener F, Peylin P, Schulz M (2010) Carbon
and nitrogen cycle dynamics in the O-CN land surface model: 2. Role of the
nitrogen cycle in the historical carbon balance. Global Biogeochemical Cycles, 24,
GB1006, doi: 10.1029/2009GB003522.
ing the last 6,000 years. Vegetation History and Archaeobotany, 17, 605–615.
Pauling A, Luterbacher J, Casty C, Wanner H (2006) Five hundred years of gridded
high-resolution precipitation reconstructions over Europe and the connection to
large-scale circulation. Climate Dynamics, 26, 387–405.
Pongratz J, Reick C, Raddatz T, Claussen M (2009) Effects of anthropogenic land
cover change on the carbon cycle of the last millennium. Global Biogeochemical
Cycles, 23, GB4001, doi: 4010.1029/2009GB003488.
Ramankutty N, Evan AT, Monfreda C, Foley JA (2008) Farming the planet: 1.
Geographic distribution of global agricultural lands in the year 2000. Global Biogeochemical Cycles, 22, GB1003..
Ramankutty N, Foley JA (1999) Estimating historical changes in global land cover:
croplands from 1700 to 1992. Global Biogeochemical Cycles, 13, 997–1027.
Robertson CJ (1956) The expansion of the arable area. Scottish Geographical Journal, 72,
1–20.
Rudel TK, Coomes OT, Moran E, Achard F, Angelson A, Xu J, Lambin E (2005) Forest
transitions: towards a global understanding of land use change. Global Environmental Change, 15, 23–31.
Schulze ED, De Vries W, Hauhs M, Rosén K, Rasmussen L, Tamm C-O, Nilsson J
(1989) Critical loads for nitrogen deposition on forest ecosystems. Water, Air, and
Soil Pollution, 48, 451–456.
Sitch S, Smith B, Prentice IC et al. (2003) Evaluation of ecosystem dynamics, plant
geography and terrestrial carbon cycling in the LPJ dynamic global vegetation
model. Global Change Biology, 9, 161–185.
Sitch S, Cox PM, Collins WJ, Huntingford C (2007) Indirect radiative forcing of
climate change through ozone effects on the land-carbon sink. Nature, 448, 791–
U794.
Six J, Conant RT, Paul EA, Paustian K (2002) Stabilization mechanisms of soil organic
matter: implications for C-saturation of soils. Plant and Soil, 241, 155–176.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Figure S1. Extending the KK10 dataset to year 2000.
Figure S2. Climate-driven changes in soil and vegetation
carbon.
Figure S3. Measured soil carbon ages in Europe.
Figure S4. Dis-equilibrium of European carbon storage at
1500 CE.
Figure S5. European forest transitions.
Table S1. Simulated carbon resulting from runs with fixed
preindustrial CO2 (~280 ppm) in living biomass, soil and litter, and total carbon in terrestrial ecosystems of Europe at
1500, 1850, 1950, and 2000 in LPJ-DGVM runs with fixed climate and reconstructed paleoclimate.
Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied
by the authors. Any queries (other than missing material)
should be directed to the corresponding author for the
article.
© 2011 Blackwell Publishing Ltd, Global Change Biology, 18, 902–914